eLISA Data Processing Centre

1. eLISA Data Processing Centre Volker Beckmann APC, Francois Arago Centre A. Petiteau, E. Porter, G. Auger, E. Plagnol, P. Binétruy

2. Data Processing Centre • Thinking about eLISA data analysis • The data stream of eLISA • Ground segment structure of eLISA • Tasks of the Data Processing Centre (DPC) • Structure of the DPC V. Beckmann

3. Thinking about eLISA data analysis • eLISA is a large and important mission • Data have to be easy to access, understand, analyse, display • Community appreciates real-time information (risk?) • Colleagues want the possibility to verify • Make all data public (raw to high level) • Provide (open source) s/w which allows community to re-do our analysis -- and potentially include own algorithms • Speak their language: astronomers like fits format, catalogs in their formats, easy access through established data bases • Listen to wishes and ideas from the larger community in order to achieve maximum support for our ambitious mission Photo: Ted Carnaham

4. Data Stream • Omni-directional observations, no ‘pointed observations’, no observers • The accumulated data provide the best sensitivity and include all information • Level 0: raw telemetry • Level 1: fully calibrated and corrected Time Delay Interferometry (TDI) data • Level 2: source identification, parameter extraction, model parameters and probability density function of sources. Transient identification • Level 3: source catalogue, (astro)physical source parameters, strain time series • Use different algorithms in the production of Level 2/3 data • Regular (e.g. weekly) reprocessing of the accumulated data stream V. Beckmann

5. Data Stream • The data volume is small, the processing demand is large • 5 Hz measurements • 30 channels ‘mother’ spacecraft, 14 channels on each of the daughter spacecrafts • Housekeeping data (at various frequencies) • 11 kbs  << Gbyte per day • Few Gbyte per day (incl. everything L0-L3)

6. Ground Segment Responsibility for data processing shared between MOC/SOC (ESA), IOTs (member states) and Data Processing Centre (member states)

7. Tasks of the DPC • Data Processing Centre (DPC) provides two analysis chains: • Quick look analysis (QA) and standard analysis (SA) pipeline. • QA: simplified version of SA. To run on a daily basis on the incoming data stream. Visualize data. Identify potential problems (beyond from what MOC/SOC) can spot. • SA: run full analysis on the accumulated data once a week • Integrate algorithms developed by the eLISA consortium members • Develop, test, maintain, document s/w chains • Distribute and provide user support for the SA pipeline • Report Level 2/Level3 results back to SOC for potential event announcements to the eLISA consortium/community. • Ensure mature level of s/w applied in the DPC • “Listen” to eLISA consortium, stay flexible

8. Structure of the DPC • Data Processing Centre (DPC) does not define the algorithms, does not (exclusively) provide scientific interpretation. • Changes in algorithms and in definition of products are to be approved by a Change Control Board (CCB) of the eLISA consortium. • Technical changes and quality control (version control, unit testing, etc.) handled locally at the DPC • Though tasks at DPC are not ‘scientific’ deep understanding of eLISA science necessary at DPC management level • Several member states can contribute to DPC (FTE or being in charge of certain tasks).

9. Structure of the DPC • Project scientist and project manager organise the work of the DPC and ensure contact with eLISA consortium, IOTs, and ESA (MOC/SOC) • Start early (but not too early!): 5 yrs before launch (1 year earlier for management) • Ramp up staff until launch • Important aspects: Quality control (testing!), maturity of s/w chain, be ready for changes after launch, maintain knowledge over the whole mission duration (including post mission activities) • DPC relies strongly on knowledge of the IOTs and the eLISA consortium • Hardware costs: mainly computing (~1,500 CPU during science operations)

10. FTEs at Data Processing Centre Launch

11. Data Processing Centre before launch

12. Data Processing Centre after launch

13. François Arago Centre • Project at APC, located across the street • Cluster, 640 CPU, 100 kW cooling, 100 TByte disk • 10 Gbit/s connection • 2 video conference rooms • Concurrent Design Facility (used e.g. for eLISA/NGO) • 2 meeting rooms • Offices on demand • Hard- and software support • Projects: LISA-PF, Euclid, CTA, INTEGRAL, HESS…

14. The DPC as member state contribution • Providing the DPC by ESA member states, reduces the costs for ESA • I.e. the more tasks on data analysis/storage/distribution/support we as consortium can provide, the lower the costs to ESA • This has been done before: e.g. in INTEGRAL, the science data centre is considered as “5th instrument” of the mission (incl. specific consortium for the SDC, I.e. with SDC under control of the member states). • Competing with ESAC • CNES starts a phase-0 study (approved today!) concerning the eLISA data processing centre requirements, risks, options, etc. • Kick-off meeting in December 2012, phase 0 to start January 2013, lasting one year.

15. Summary • eLISA has to provide excellent service to the whole community, before, during, after the mission • Open to ideas, requests, wishes, concerns • DPC processes L1 to L3 data • Small data volume, large processing needs • Quick look and standard analysis (simulator not included) • Development, integration, testing, maintenance, documentation, user support • Algorithm development by eLISA consortium (not DPC), knowledge about instruments with IOTs • Start 2017 with 2 FTE management, 2018 first s/w integration activities • 58 FTE*years until launch • French community is ready to provide the Data Processing Centre services to the eLISA consortium! • See also Beckmann et al. 2012, “The eLISA/NGO Data Processing Centre”, ASP Conf. Series 467 V. Beckmann